BEVMAPMATCH: Multimodal BEV Neural Map Matching for Robust Re-Localization of Autonomous Vehicles

Shounak Sural, Ragunathan Rajkumar · 2026

Read BEVMAPMATCH: Multimodal BEV Neural Map Matching for Robust Re-Localization of Autonomous Vehicles by Shounak Sural, Ragunathan Rajkumar (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-25963.txt.

Description

Sural and Rajkumar present BEVMapMatch, a way for self-driving cars to localize without satellite signals. GPS can fail or degrade in tunnels, cities, and bad weather. Their system fuses lidar and camera data into bird's-eye view segmentations. A cross-attention search then matches those maps against patches of a known map. The best patch is aligned in detail for a global position fix. Stacking several frames sharpens the result further. The authors report a Recall at one meter of 39.8 percent. That is nearly twice the best rival method.

Catalog record

CREATORShounak Sural, Ragunathan Rajkumar
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38300
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.25963

Browse the full library · About the Tower of Babel